Characteristics of the Disastrous Debris Flow of Chediguan Gully in Yinxing Town, Sichuan Province, on August 20, 2019

Li Ning (  [email protected] ) School of Emergency Science, Xihua University Tang Chuan State Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu Zhang Xianzheng State Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu Chang Ming State Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu Shu Zhile School of Emergency Science, Xihua University Bu Xianghang State Grid Sichuan Electric Power Research Institute,Chengdu, Sichuan

Research Article

Keywords: debris ow, disaster characteristics, Wenchuan earthquake, drone measurement, sediment supply conditions

Posted Date: August 25th, 2021

DOI: https://doi.org/10.21203/rs.3.rs-798905/v1

License:   This work is licensed under a Creative Commons Attribution 4.0 International License. Read Full License

1 Characteristics of the disastrous debris flow of Chediguan gully in Yinxing 2 town, Sichuan Province, on August 20, 2019 3 LI Ning①,⁎, TANG Chuan②, ZHANG Xianzheng②, CHANG Ming②, SHU Zhile①, BU Xianghang③ 4 (①School of Emergency Science, Xihua University, Chengdu 610039 China,

5 ②State Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu,61005

6 China,

7 ③State Grid Sichuan Electric Power Research Institute, Chengdu, Sichuan 610072, China)

8 ⁎ Corresponding author at: School of Emergency Science, Xihua University, No.9999 Hongguang Rd, Chengdu, 9 Sichuan 610039, China. Email: [email protected].

10 Abstract: On August 20, 2019, at 2 a.m., a disastrous debris flow occurred in Chediguan gully in Yinxing town, 11 China. The debris flow destroyed the drainage groove and the bridge at the exit of the gully. In addition, the debris 12 flow temporarily blocked the Minjiang River during the flood peak, flooding the Taipingyi hydropower station 200 13 m upstream and leaving two plant workers missing. To further understand the activity of the debris flow after the 14 Wenchuan earthquake, the characteristics of this debris flow event were studied. Eleven years after the Wenchuan 15 earthquake, a disastrous debris flow still occurred in the Chediguan catchment, causing more severe losses than 16 those of earlier debris flows. In this paper, the formation mechanism and dynamic characteristics of this debris flow 17 event are analysed based on a drone survey, high-definition remote sensing interpretations and other means. The 18 catastrophic debris flow event indicates that debris flows in the Wenchuan earthquake area are still active. A large 19 amount of dredging work in the main gully could effectively reduce the debris flow risk in the gully. In addition, it 20 is also important to repair or rebuild damaged mitigation measures and to establish a real-time monitoring and early 21 warning system for the high-risk gully. 22 Key words：debris flow, disaster characteristics, Wenchuan earthquake, drone measurement, sediment supply conditions

23 0 Introduction 24 At 2:28 pm on May 12th, 2008, an MS8 earthquake struck Wenchuan County in Sichuan Province (“5.12” 25 earthquake), China. The earthquake induced more than 56,000 landslides in a mountainous area of 41,750 km² (Dai 26 et al. 2011). Most of the landslides were distributed in hills and gorges, and the landslides converted into frequent 27 debris flows under heavy rainfall after the earthquake (Chang et al. 2017, Fan et al. 2018). During the 11 years that 28 have passed since the Wenchuan earthquake, the long-term debris flow activities occurring after the earthquake have 29 caused great threats to local residents. Among them, the most serious debris flows occurred on August 14, 2010, in 30 Longchi town and Yingxiu town and on July 10, 2013, in Wenchuan (Chang et al. 2014, Tang et al. 2011a). Some 31 scholars have pointed out that strong debris flow activities after earthquakes will last 10~15 years or even 30 years 32 (Tang et al. 2011b, Chen et al. 2019). This concept is well supported by the mass debris flow that occurred in 33 Wenchuan County on August 20, 2019. 34 From 0:00 to 7:00 on August 20, 2019, the cumulative rainfall in Wenchuan County reached a maximum of 65 35 mm, resulting in a number of debris flows and flash floods and affecting more than 10 townships in the county. 36 Among them, the disasters in Miansi town, Sanjiang town, and Yinxing town were the most serious. According to 37 the information released by the Wenchuan County People's Government, the “8.20” disaster damaged many roads, 1

38 pieces of communication equipment, farmlands and houses. The local government transferred a total of 48,000 39 people and resettled 2,505 people. As of September 5, 2019, the debris flows and flash floods caused 16 deaths and 40 25 missing people, resulting in direct economic losses of approximately 3.408 billion yuan. According to the survey, 41 the losses were mostly caused by flash floods. The debris flow events mainly caused damage to the bridges and 42 infrastructure along the Minjiang River, hindering the transportation, power supply and communication in the 43 disaster area and increasing the difficulty of rescue but not causing casualties. 44 On August 20, 2019, at 2 a.m., debris flow occurred in Chediguan gully of Yinxing Town, with a volume of 45 63.8×104 m3. The debris flow poured out along drainage channels and destroyed the drainage groove located at the 46 ditch as well as the G213 Taiping Middle Bridge. According to the photos taken by the UAV, the debris flow fan was 47 300 meters long, 260 meters wide, and the average depth is 8 meters. The volume and area of the deposition area 48 were 63.8×104 m3 and 7.9×104 m2, respectively (Figure 1). The debris flow front blocked Minjiang River and caused 49 the water level to rise, flooding the active dam area and leaving two people missing. Debris flows broke out several 50 times in Chediguan gully after the “5.12” earthquake (Figure 1 and Table 1). 51 As Chediguan gully is located along on the only routes of the Du-Wen highway and 213 national road, 52 Chediguan debris flow disasters always threaten the roads and passing vehicles. After the completion of mitigation 53 measures in 2011, the debris flow activity in Chediguan gully was significantly reduced, and the debris flow risk 54 gradually became ignored. The debris flow event in 2019 brought the Chediguan gully back into our view and left us 55 with many questions. The most important question involves the cause of the disaster. What we need to know is why 56 the Chediguan gully, under the defence of mitigation measures, can still break out in such a large-scale debris flow 57 10 years after the earthquake and cause such serious damage. Based on this concern, we investigated the debris flow 58 disaster that occurred on August 20, 2019, and analysed the formation mechanism and dynamic characteristics of 59 this debris flow event by means of a drone survey, high-definition remote sensing interpretation and other means. 60 This paper can provide a reliable scientific basis for the management of the Chediguan catchment and enrich the 61 understanding of post-earthquake debris flows.

62 63 Fig 1 Mizoguchi photos of Chediguan gully after the debris flow in 2011 and 2019

64 1 Study Area 65 Chediguan gully is located in Yinxing town, Wenchuan County, Sichuan Province, China (Figure 2). The 66 coordinates of its mizoguchi are 103°29’2”E and 31°13’23.4”N. The flow direction of the gully is from west to east, 67 and the gully covers an area of 16.49 km2. The main ditch is approximately 6.2 km long with an average 68 longitudinal slope of 302‰. Maximum altitude 2940 m, minimum altitude 1065.5 m, and the height difference 2

69 1574.5 m. The study area is approximately 18 km away from Yingxiu town and approximately 35 km away from 70 Wenchuan County. The Yingxiu-Wenchuan Expressway runs through the ditch along the tunnel and provides 71 convenient transportation. The geographical location of the gully is shown in Figure 2.

72 73 Fig.2 Location map of the study area 74 Chediguan gully belongs to middle and high mountainous areas, with steep overall terrain and free face 75 development in the catchment. The drainage system in Chediguan gully is developed, and there are a total of 9 76 tributaries along the main channel. All the tributaries are steep, and their lengths and watershed areas are quite 77 different from each other. Compared with the tributaries on the left bank, the tributaries on the right bank are shorter, 78 their areas are smaller, and their slopes are steeper. The developed drainage system can not only effectively collect 79 rainfall but can also promote the migration of source material in the basin to the main channel, thus promoting the 80 formation of debris flows. The drainage system and a topographic map of Chediguan gully are shown in Figure 3. 81 The catchment is mainly formed by steep slopes, which are beneficial for accumulating rainfall and causing 82 landslides. Among the whole catchment, gently sloping lands (<25°) accounting for 11.52% of the total area, which 83 distributed in the downstream area. Steep lands (25°– 35°) accounting for 8.79% of the total area, and acutely steep 84 lands (≥35°) accounting for 79.69% of the total area. The main channel slope is steeper at and above the intersection 85 with the 6th tributary (Feixianyan gully) but gradually slows below this confluence.

3

86 87 Fig. 3 Drainage system and topography map of the Chediguan gully 88 The geological environment of the study area is complicated. It is located in the middle of the 89 Longmenshan-Huaxia tectonic belt with a NE direction of 40°~50°, approximately 14 km away from the Yingxiu 90 fault on the southeast side and approximately 15 km away from the Mao-Wen fault on the northwest side. The 91 lithology of the strata that are exposed along the main channel mainly include biotite granite (γ42) of the Proterozoic 92 Jinning period, Quaternary alluvial-diluvial deposits (Qal+pl4) and colluvial deposits (Qc+dl4) and seismic deposits 93 (Qc4) resulting from fragmented stones, soils, and debris flow sediments (Qsef 4). Affected by the Maowen fault, the 94 biotite granite (γ42) is severely broken, the thickness of the weathered layer is large, and multiple structural tectonic 95 fissures are developed in the rock stratum, easily forming collapses and landslides. In the middle and lower parts of 96 the basin, secondary faults are distributed along the ditch that belong to the secondary fault structure of the 97 Longmenshan central fault, and the two sides of the main channel have critically collapsed. 98 There are frequent earthquakes in the study area. According to the current records, there have been 8 major 99 earthquakes and many strong earthquakes of magnitude 5 or higher. For example, the Diexi MS7.3 earthquake 100 occurred on August 25, 1993, and the Wenchuan MS8.0 earthquake occurred on May 12, 2008. Based on “the 101 ground motion parameter zoning map of the Wenchuan earthquake” (GB18306-2001), the study area belongs to a 102 high-intensity seismic region (VIII degrees) with a peak ground acceleration of 0.2 g and a characteristic period of 103 the seismic response spectrum of 0.4 s. Frequent earthquakes result in loose rock and soil structures in the ditch and 104 abundant reserves of original provenance. According to a survey, the total amount of loose source materials in 2011 105 reached 742.679 ×104 m³, and the reserves were 194.525×104 m³ in volume (Zhang 2016). The abundant source 106 material provides favourable conditions for debris flows. 4

107 The study area is located in a subtropical humid climate zone and is a concentrated area of heavy rainfall in the 108 Minjiang River Basin. The average annual rainfall in this area is 1253.1 mm, the maximum annual rainfall is 1688 109 mm, and the minimum annual rainfall is 836.7 mm. The maximum continuous rainfall for 4 months (June-September) 110 is 853.2 mm, comprising 68.2% of the annual rainfall. The rainfall in the survey area is abundant and concentrated 111 and can meet the hydrodynamic conditions required to stimulate debris flows. 112 According to the investigation and interview, this gully belongs to an old debris flow gully, and there have 113 been four mudslides in recorded history (Table 1). One mudslide occurred in the year 1952 (He 2015), but its scale 114 and the damage caused by it have remained unascertained for a long time. The second mudslide occurred on the 115 evening of August 13, 2010 (Tu et al. 2013). When the accumulated rainfall reached 102.8 mm and the hourly rain 116 intensity reached 14.2 mm/h, debris flows broke out in the tributaries on both sides of the gully, but no debris flow 117 occurred in the main channel. The third mudslide occurred on July 3, 2011. Affected by heavy rain, a debris flow 118 comprising approximately 1.5×104 m3 of material was washed down from upstream and accumulated in the 119 downstream main channel, destroying the drainage canal on the upper side of the Ying-Wen Highway and some 120 mechanical equipment. The fourth mudslide occurred on July 20, 2011. Heavy rainfall caused a large-scale debris 121 flow in the early morning. The total amount of rushing material reached 10×104 m3, causing the lateral 122 displacement of the G213 bridge offset by 12 cm and interrupting traffic for 2 d. At the same time, more than 1/3 of 123 the Minjiang River was blocked by debris. 124 Table 1 Typical debris flow events of Chediguan Gully

Barrier dam parameters (m) Magnitude Time Rainfall (mm) Impact Data source Max length Max width Average depth (×104 m³)

No casualties or economic damage were 1952 —— —— —— —— —— —— (Zhang 2016; He 2015) caused 19.5 55.3 2010-8-13 —— —— —— —— Debris flow broke out in the tributary (Tu et al. 2013) （1h） (72h) The debris flow drainage canal under 14.2 102.8 (He 2015; Fan et al. 2011-7-3 —— —— —— 1.5 construction and some mechanical (1h) (72h） 2019) equipment were destroyed

G213 bridge body offset 12cm, the 40 300 tunnel under construction in the ditch (Fan et al. 2019; He 2011-7-20 100 348 15 52.9 (1h) (72h) partial collapse, the debris flow body 2015) temporarily blocked the Minjiang River

Two people are missing. Two Bridges, one Field investigation and 17.8 43.2 2019-8-20 300 260 8 63.8 house and about 650m country road were the Wenchuan County (1h) (24h) destroyed Meteorological Bureau

125 2 Data and Methodology 126 2.1 Field investigation

127 Through interviews with local residents to understand the Cheguan gully debris flow history, as well as debris 128 flow characteristics and other basic information. A 1:200 topographic map of the deposition area and a 1:200 cross 129 (longitudinal)-section map of the downstream channel (C1–C6 (C7) in Figure 5A and B) were created by UAV 130 mapping method. As shown in Fig. 4, debris flow samples were collected at three sites (S1–S3) from the 131 downstream channel and debris fan for sieving and particle gradation analysis (Figure 4c). Field and laboratory dry 132 sieving tests were conducted following the British Standards. mostly composed of gravel and cobblestones. At the 133 same time, field screening tests were carried out on the site of exposed sand and gravel. After recycling, weighing

5

134 and drying, the materials were divided into 2-5, 5-10, 10-20, 20-60, 60-100, 100-150, 150-200 and greater than 200 135 mm. Fine materials (less than 2 mm in diameter) were further tested using laboratory dry screen tests.

136 2.2 Drone aerial photography and measurement

137 We used a drone to photograph and map the areas below the No. 4 dam at Chediguan gully on December 10, 138 2017, April 21, 2019, and September 28, 2019, as shown in Table 2 and Fig. 4(a/b/c). The accuracy of the terrain 139 data obtained by the drone is as high as 1 cm, and the data can be used to accurately analyse the distribution of the 140 source material and create a digital surface model (DSM). The DSMs obtained by the drone representing three 141 periods were used to analyse the movement characteristics of sediments in the middle and lower reaches of the 142 channel. Then, the differences in the DSMs between September 28 and April 21, 2019 was obtained by using the 143 spatial analysis function of ArcGIS; these differences can reflect the topographic changes that occurred before and 144 after the "8.20" debris flow (Fig 4d).

145

146 Fig. 4 Field investigation sketch map. (a) Drone image taken on Dec. 10, 2017; (b) Drone image taken on Apr. 21, 2019， 147 the distribution of sampling sites and cross (longitudinal)-section map are also noted; (c) Drone image taken on Sep. 28, 148 2019, the distribution of sampling sites and cross (longitudinal)-section map are also noted; (d) Raster image obtained by 149 subtracting the DSM of April 21, 2019 from the DSM of September 28, 2019.

150 The data showed that the "8.20" debris flow was mainly eroded from the middle and lower reaches of the 151 channel, with an average erosion depth of 1.66 m, a total erosion depth of 1.64×104 m³ and a total erosion amount 152 of 39.59×104 m³ (Figure 4d). A large number of large-scale landslides and tributary debris fans on both sides of the 6

153 main channel provided abundant loose material replenishment for the “8.20” debris flow, increasing the scouring 154 force of the debris flow; thus, the drainage canal located before the No. 1 dam was completely destroyed. 155 Debris flows were mainly deposited in the mizoguchi of the gully, and a total volume of 63.83×104 m3 was 156 deposited compared with the terrain surveyed on April 21, 2019; this number represents the accumulation amount 157 of the "8.20" debris flow. As a large number of deposits congested the river, the riverbed of the Minjiang River was 158 uplifted, and the river surface increased by 3.5 m on average before the outbreak. 159 Table 2 Data used in this study Data Type Source Acquisition date Resolution Application 2008/9/9 Google Earth Images 2011/4/26 0.5m Generating provenance 2015/4/15 inventories Images 2019/8/7 Thematic Mapper 30m 2019/8/24 2019/4/21 Drone aerial photography 1cm Topographic survey 2019/9/28

160 2.3 Multi-temporal source materials inventories

161 Determining the provenance is the basic condition for studies of debris flows. Therefore, the key to 162 determining the cause of a debris flow is to thoroughly determine the evolution law and migration form of the 163 provenance of the associated debris. Based on this concern, a digital stereoscopic image interpretation was used to 164 map the source material inventories (Table 2, Fig 5). A resolution of 0.8 m on September 9, 2005, was used to 165 reflect the source material conditions before the earthquake. Two Thematic Mapper (TM) satellite images with 166 30-m resolutions from the website of the US Geological Survey were used to analyse and compare the formation 167 conditions of debris flows before and after the debris flow broke on 20th Aug 2019.

168 169 Fig.5 digital stereoscopic images used in this paper 170 The source materials were classified by their mass movement types (Tang et al. 2016) (Fig. 6). We 171 differentiated the following mass movement types: fall, slide, flow, fall-slide, slide-flow, and slide-fall. In the case 172 of fall, materials fall from steep cliffs, with little additional displacement. Bedrock can be seen very clearly in the 173 scarp area, and the accumulation area often tends to be cone-shaped. Slide-type movements are characterized by 174 clear back scarps and the identification of a sliding mass that is either translational or rotational in form. Flow-type 7

175 movements are mostly confined to channels and occur mostly as debris flows. Fall-slide, a combination of falling 176 and sliding, can be observed when fall-type movement occurs on a steep slope and the deposits slide down further 177 during or after deposition. Slide-fall movements initiate as a slide on top of a steep cliff, and the slide materials 178 subsequently fall over the cliff. A very common combination of landslide types is the slide-flow type, wherein the 179 source material areas of a debris flow are formed by one or more slide-type movements.

180 181 Fig.6 Examples of source materials movement types. 182 2.4 Parameter calculating 183 The unit weight is one of the most important parameters of a debris flow. It not only represents the 184 concentration of the debris flow but also the necessary data for calculating the dynamic parameters of the debris 185 flow. There are many methods for measuring the unit weight of a debris flow, among which the most accurate 186 method involves field sampling and measurements. In addition, formulation methods and statistical formulas are 187 also commonly used (Yu 2008). Due to the great subjective influence of witnesses, the accuracy of assessments of 188 the grouting method cannot be guaranteed. Moreover, no suitable witnesses were found in the study area. Therefore, 189 this paper chooses to use the statistical formula method to calculate the unit weight of the “8.20” debris flow (Yu 190 2008). 191 This method, through the statistical analysis of the debris flow using three characteristic particle sizes, 192 represents the coarse particle size, the particle size of fine particles and the particle size of clay particles (2 mm, 193 0.05 mm and 0.005 mm, respectively) as well as the percentages of their relationships with the total debris flow unit 194 weight and the correlations between the percentages of coarse and fine particles greater than 2 mm and less than 195 0.05 mm and the bulk density of the debris flow, as shown in formula (1) (Yu 2008). 196 (1) 0.35 퐷 05 2 푣 0 197 Where P05 is the percentage of fine particles훾 less= 푃 than푃 0.05훾 + mm 훾 (in decimal); P2 is the percentage of coarse 8

3 198 particles larger than 2 mm (in decimal); γv is the minimum unit weight of viscous debris flow, =2.0 g/cm ; γ0 is the 199 minimum unit weight of the debris flow, = 1.5 g/cm3. 200 A summary of the parameters of the “8.20” Chediguan debris flow is shown in Table 3. The particle size 201 distributions of debris flows are shown in Figure 7.

202 Table 3 Parameters summary of debris flow samples

γD Difference with 'S#-1'(%) Time No. Site P05 P2 R /g/cm³ P05 P2 R γD

S1 behind No.2 dam 0.044 0.654 0.53 1.938 50.00 -20.03 48.23 1.34 2019/ S2 behind No.1 dam 0.031 0.559 0.79 1.831 90.32 -26.83 47.97 8.65 9/28 S3 debris fan 0.005 0.72 0.39 1.725 —— —— —— ——

2019/ S1-1 behind No.2 dam 0.022 0.785 0.27 1.913 —— —— —— —— 4/21 S2-1 behind No.1 dam 0.003 0.709 0.41 1.686 —— —— —— ——

203 204 Fig. 7 Particle size distributions of debris flow samples Fig. 8 Distribution of hourly and accumulated 205 rainfall on August 18–21, 2019.

206 3 Results and Discussions

207 3.1 Disaster characteristics 208 3.1.1 Triggering rainfall 209 Strong earthquakes cause the porosity of a source material to increase and become looser. At the same time, 210 the source material becomes more prone to instability under the influence of rainfall and can then transforms into a 211 debris flow. Therefore, the hydrodynamic conditions required for a debris flow are lower after an earthquake, as is 212 the rainfall threshold (Zhou and Tang 2014). Tang and Liang (2008) indicated that compared with the situation in 213 Beichuan County before the earthquake, the critical hourly and cumulative rainfall decreased by 25.4-31.6 % and 214 14.8-22.1 %, respectively. 215 According to the measured dataset at the rainfall monitoring station of Taoguan Village in Yinxing Township 216 (Fig. 1/8), rainfall began in the study area at 4:00 on August 19, 2019. The accumulated rainfall on that day was 217 23.1 mm. The accumulated rainfall before the debris flow broke out at 2 o'clock on August 20, 2019, was 20.1 mm; 218 that is, the accumulated cumulative rainfall in the complete debris flow reached 43.2 mm. 219 The rainfall that eventually induced the debris flow appeared from 2:00 to 3:00 am on August 20, 2019. The 220 maximum rainfall rate was only 17.8 mm/h, representing a slow-rising rain type. It should be noted that 48 h of 9

221 rainfall also broke out in the area a month earlier, with a total rainfall of 45.2 mm and a maximum rainfall rate of 222 13.6 mm/h, which translated sediment from the tributaries and slope to the main channel and contributed to the 223 debris flow on August 20, 2019. The typical debris flow events that have occurred in the past in Chediguan Gully 224 are shown in Table 1. The accumulated rainfall amounts in the early stage of the mudslides that occurred on August 225 20, 2010, November 4, 2011, and November 20, 2011, were 55.3 mm (72 h), 102.8 mm (72 h), and 300 mm (72 h), 226 and the excitation rain intensities were 19.5 mm/h, 14.2 mm/h, and 40 mm/h, respectively. Compared with past 227 typical events, the pre-accumulated rainfall amount before the “8.20” debris flow was smaller than those in 228 previous years, and the critical rainfall intensity decreased by 8.7%~55.5%. 229 What puzzles us is why the scale and damage of the "8.20" debris flow were higher than ever under less 230 rainfall. This phenomenon also shows that rainfall is not the only cause of debris flow events. Our findings show 231 that there may be three reasons for this phenomenon. More source material in the main channel is the first and most 232 important reason. The investigation found that after the debris flow broke out on July 20, 2011, the No. 1 to No. 4 233 dams were filled with sediments from the tributaries and slopes. The second reason is the rainfall that occurred on 234 July 19-21, 2019. This rainfall event transported the source materials in the tributaries and slopes into the main 235 channel and at the same time gave these materials a higher moisture content, making them easier to be initiated into 236 a debris flow. The third reason is the difference in the rainfall data. Our rainfall data are taken from lower altitudes 237 than that of the study site, while debris flows usually form at higher elevations. Studies show that higher rainfall 238 occurs at higher elevations (Zhou et al. 2019, Tzimopoulos et al. 2018). Therefore, our rainfall data cannot 239 represent the actual rainfall that triggered the "8.20" debris flow. We will continue to explore the possibilities for 240 these reasons in the following chapters.

241 3.1.2 Sediment supply conditions

242 The multiphase source distribution of Chedaiguan gully is shown in Fig. 9. The changes in source materials 243 that occur with different movement types are shown in Table 4 and Fig. 10. As shown in Table 4, before the 244 earthquake, the total area of source materials developed in Chediguan gully was only 2.86×104 m2, which increased 245 to 473.8×104 m2 in 2011. After the earthquake, from 2011 until August 2018, the number of source materials in the 246 study area decreased by 83 to a total of 123. The total source material area increased from 473.8×104 m2 to 247 265.6×104 m2. The total amount of source materials in Chediguan gully after the “5.12” earthquake shows a 248 continuously decreasing trend, while source materials that underwent different movement types show different 249 change rates. 250 The results show (Fig. 10) that the total area of the flow-moving source materials increased by 6.04% from 251 2011-2015, while the other areas decreased considerably. Between 2015 and August 7, 2019, the total areas of 252 source materials moving as slide-flow, fall-slide, and flow types increased by 9%, 7.23% and 11.39%, respectively, 253 while the total source material areas moving as slide-fall, slide, and fall types decreased by 14.06%, 30.79% and 254 26.14%, respectively. After the debris flow occurred on August 20, 2019, the number of source materials in 255 Chediguan gully increased to 127. The total source material area increased to 281.2×104 m2. Among the source 256 materials, the areas of source materials moving as slide-falls and flows increased the most, with growth rates of 257 16.36% and 31.82%, respectively; these were also the main source material of the "8.20" debris flow sediment 258 supply. 259 The observed changes in the data indicate that the source materials after the “5.12” earthquake evolved in 260 different ways. First, the slide-flow and slide types are the main movement types of the provenance material. In 10

261 2011, the source materials moving in these two movement types accounted for 41.75% and 20.05% of the total, 262 respectively, while the data on August 24, 2019 showed 35.38% and 33.85% of the total. Second, the source 263 materials were continuously moved to the lower region in the form of slide-flows, fall-slides, and flows under the 264 effects of runoff erosion and gravity. The third finding is that the source material in the watershed mostly moves as 265 the flow movement type, and its area is expanded by the formation of lateral and downward erosion (Fig. 4); this 266 erosion was the main cause of the "8.20" debris flow. At present, the provenance material in Chediguan gully is 267 mainly distributed in the main gully and moves in the form of slide-flows and flows. Therefore, dredging work in 268 the main channel is necessary to reduce the possibility of debris flows.

269 270 Fig.9 Multiphase provenances distribution map of Chediguan Gully

271 272 Fig.10 Change of source materials with different movement types based on multiple phases of remote sensing 273 images 274 275 11

276 Table 4 Statistics of the source materials inventories before and after the earthquake. Movement types Total Fall Slide Flow Fall-slide Slide-flow Slide-fall Time number Area number Area number Area number Area number Area number Area number Area (×104m2)

2005/9/9 0 0 6 1.83 0 0 0 0 4 1.03 0 0 10 2.86 2011/4/26 27 27.5 83 122.5 7 28.2 25 95 62 197.8 2 2.8 206 473.8 2015/4/15 11 24.1 90 129.9 5 23.7 9 24.9 41 91.1 1 6.4 157 300.1 2019/8/7 9 17.8 63 89.9 5 26.4 9 26.7 36 99.3 1 5.5 123 265.6 2019/8/24 10 18.2 67 95.2 5 34.8 9 26.7 35 96.4 1 6.4 127 277.7

277 3.1.3 Deposition characteristics

278 In the '8.20' event, debris flow material was transported to the gully mouth and formed a large debris flow fan. 279 (Fig. 11). The fan was 300 m long and 260 m wide, with an average depth of 8 m (Fig 4). The aerial photos show 280 that the debris flow fan area was 3.1×104 m2 and the volume was about 63.8×104 m3. The debris flow destroyed the 281 bridge at the mizoguchi, destroyed many houses at the opposite bank of the Mingjiang River, and finally silting into 282 a debris fan. Figure 12 shows the significant changes that occurred in the middle-lower gullies and gully mouth 283 before and after the “8.20” debris flow event. The debris flow consists mainly of erosion before the C6 284 cross-section and mainly of deposition after the C6 cross-section. The highest deposition depth at the mizoguchi 285 reached 9.5 m, the riverbed was uplifted, and nearly 1/3 of the river was buried. The debris flow blocked the 286 Minjiang River temporarily during the flood peak, causing the river to return and flood the Taipingyi hydropower 287 station 200 m upstream, leaving two plant workers missing.

288 289 Fig.11 Debris fan and barrier lake formed in the “8.20” debris flow event

290 291 Fig. 12 Longitudinal cross-section of the “8.20” debris flow event 292 To obtain the particle grading of the debris flow granules, we went to Chedgiuan gully on April 21, 2019, and 293 September 28, 2019, and obtained a total of 5 soil samples (S1, S2, S3, S1-1, and S2-1). The sampling positions are 294 shown in Figure 4(c). The sampling point S3 was located in the middle of the debris fan and can be used to 12

295 calculate the debris flow unit weight. The particle size distributions of the three sets of soil samples obtained by 296 sieving are shown in Fig. 5. The P05 of S3 is 0.005, and its P2 is 0.72. By substituting these values into formula (1), 297 it can be obtained that the unit weight of the debris flow was 1.725 g/cm, which is larger and belongs to the class of 298 sub-viscous debris flows. According to the survey conducted by the Sichuan Metallurgical Geological Exploration 299 Bureau, the unit weight of the debris flow that occurred on July 20, 2011, was 1.866 t/m3, higher than the unit 300 weight of the "8.20" debris flow. 301 Particle gradations at different locations and over different periods can visually reflect the transport processes 302 of different particles during debris flows. The data of all the samples are shown in Table 2 and Figure 7. It can be 303 seen that the P05 of the soil sample collected in the channel after dam No. 1 increased by 50%, the P2 decreased by 304 20.03%, and the earth-rock ratio increased by 48.23%, while the unit weight increased by 1.34%. The soil samples 305 collected after dam No. 2 showed similar trends, with the P05 increasing by 90.32%, the P2 decreasing by 26.83%, 306 the earth-rock ratio increasing by 47.97%, and the unit weight increasing by 8.65%. These changes in the collected 307 data show that after the debris flow events, the fine particle content measured after the dam and the soil-rock ratio 308 were greatly increased, while the coarse particle content was reduced. The parameter differences among S1, S2, and 309 S3 reflect the particle movement process during the "8.20" debris flow. The P05 values of S1, S2, and S3 were 310 0.044, 0.031, and 0.005, and the P2 values of S1, S2, and S3 were 0.654, 0.559, and 0.72, respectively. In other 311 words, the fine particle content gradually decreases from the upstream region to the debris fan, while the coarse 312 particle content increases continuously. 313 Therefore, the "8.20" debris flow was mainly formed by coarse particles, and a large number of fine particles 314 from upstream were blocked by the dams. The interception of the dam caused an increase in the soil-rock ratio and 315 soil unit weight of the sediment behind the dam. Coarse particles continued to move downstream with the debris 316 flow, thereby reducing the unit weight of the debris flow sediment. 317 3.2 Formation mechanism and dynamic characteristic

318 3.2.1 Formation mechanism 319 The previous analysis shows that the “8.20” Chediguan debris flow event had the following characteristics: (1) 320 the rainfall that occurred one month before the debris flow brought the sediment to the main channel together; (2) 321 the source material was started collectively by downcutting of the broached groove; (3) the debris flow movement 322 was mainly undercut erosion and lateral erosion in the main channel; and (4) the impact force of the debris flow 323 was extremely high. 324 As shown in Figure 10, the total areas of the provenance materials experiencing slide-flow, fall-slide, and flow 325 movement on August 7, 2019 were 9%, 7.23%, and 11.39% higher than that of 2015, respectively, and the bright 326 colours shows that these movements occurred recently. Fig. 8 shows that the concentrated rainfall that occurred on 327 July 18-20, 2019, is one of the causes of sediment movement. Heavy rainfall caused the initiation of some 328 landslides and the initiation of debris flows in some tributaries. 329 During the downward migration of the tributary debris flows, the coastal sediments were continuously washed 330 and finally entered the main channel. Within one month before the “8.20” debris flow, the sediments continued to 331 move toward the main channel, providing abundant provenance conditions for the debris flow. The previous rainfall 332 also gave the old deposits a higher moisture content, making them easier to initiate into a debris flow. According to 333 the conducted interview, the Chediguan debris flow began at approximately 2 a.m. on August 20th, 2019. Heavy 334 rainfall again induced mudslides in the tributary. The rushing material of the upstream tributary entered the main 335 channel to form a debris barrier dam at a great rate, eroded by the upstream inflow by cutting down the groove (Fig. 336 13a/b/c/e), and finally converted to a mountainous torrent and debris flow. 337 The upstream debris flow rapidly expanded downstream via forward erosion and continuously revealed 338 sediments along the way (Fig 13 a/b/c/e). The debris was like a “snowball” that became increasingly larger, 339 destroying three sand-blocking dams and bridges in the middle and lower reaches of the channel (Fig. 13 d/f/g/h). 340 Finally, the debris flow rushed into the Minjiang River, caused a short-term blockage, and then deposited and

13

341 formed a debris fan in the mizoguchi section. In conclusion, the Chediguan debris flow had an obvious chain effect, 342 characterized by the combination of heavy rain (mountainous torrent)-collapse landslide-tributary debris 343 flow-rainfall-channel erosive erosion to form the debris flow. The formation process of the “8.20” Chediguan debris 344 flow is shown in Figure 14.

345 346 Fig. 13 Characteristics maps of Chediguan debris flow broken in 20th August. (a) the collapse in front of no. 4 dam 347 due to erosion by water flow; (b) the debris fan of no.8 tributary was eroded;（c）the old deposition fan and slope 348 behind no 2 dam was eroded;（d）the no. 2 dam was destroyed;（e）channel erosion characteristics below no 1 dam ; 349 （f）the channel in front of no. 1 dam was severely down-eroded;（g）the drain groove at the mizoguchi was 350 completely destroyed;（h）two Bridges at the mizoguchi were washed away.

14

351 352 Fig. 14 Formation process sketch maps of Chediguan debris flow broken in 20th August.

353 3.2.2 Dynamic characteristic

354 (1) Flood peak discharge 355 Firstly, the flood peak discharge (QP) of “8.20” Chediguan debris flow should be determined for the following 356 calculation of debris flow peak discharge (Qc). The flood peak discharge (QP) can be calculated by (1) (Liu et al. 357 2015): s 358 Q==0.278 iF 0.278 F (1) p  n 359 where F is the catchment area; ψ is the runoff coefficient of flood peak; S is the maximum rainfall in an hour and 360 equal to 17.8 mm/h; τ is the runoff confluence time of the rainstorm; and n is the attenuation index of the rainstorm. 361 Figure 8 showed the rainfall distribution of hourly and accumulated rainfall on August 18–21, 2019. ψ, τ, and n can 362 be determined by the following empirical equations (Liu et al. 2014).

μ 363 ψ=1- τn (2) S n n S 1-n -0.19 364 μ=(1-n)n1-n( ) =3.6F (only applicable to Sichuan Province) (3) hn

365 4 (4) 0.383 4−푛 1/4 366휏 = [푚푆m=0.318θ/휃] 0.204 367 (5)

L 368 θ= J1/3F1/4 369 (6)

H 370 n=1+1.285( log 6p ) H24p 371 (7) 372 where m is the runoff confluence parameter; θ is the 373 catchment characteristic parameter; L is the channel length; 374 J is the longitudinal slope of the channel; μ is the runoff 375 yield parameter; H6p is maximum rainfall in 6 hours and 376 equal to 60.2 mm/h; H24p is maximum rainfall in 24 hours 377 and equal to 60.2 mm/h. 378 （2）Debris flow dynamic parameters 379 For the following discussion on dynamic properties and 380 hazard predictions, some dynamic parameters, including the

381 debris flow velocity (Vc), peak discharge (Qc), the total 15

Fig. 15 The measured cross-section (C1–C7)

382 volume of one debris flow (Qt) and the debris flow impact force (FC), need to be determined. A total of 6 sections 383 were selected for which to calculate these parameters. The distribution of these sections is shown in Figure 1, and 384 their terrain profiles are shown in Figure 15.

385 The debris flow velocity (Vc) can be calculated by (Tang et al. 2011): 1 386 VHI=2/3 1/2 (8) CC()C nC

387 where Hc is the hydraulic radius of the debris flow, can be replaced by the average deep mud; Ic is the hydraulic 388 slope of debris flow (J), can be replaced by the longitudinal grade of channel; and nc is the roughness coefficient 389 and determined from an assignment table which is based on the debris flow fluid characteristic and channel 390 condition (Liu et al. 2014).

391 The debris flow peak discharge (Qc) can be calculated by two methods (Liu et al. 2014): 392 stormwater method: QQDc=()1 + p  c (9) 393 morphological survey method: Qc= (10) 394 where Dc is the debris flow blockage coefficient. Generally, the Dc is divided into three interval according to the 푊퐶 ∗ 푉푐 395 blockage degree: 1.0-2.5 (minor), 2.5-3.5 (normal), 3.5-4.5 (serious) and 4.5-5.5 (very serious). Based on the field 396 investigation, the Dc of C1-C6 cross-section of Chediguan Gully is considered as 1.2-2.5 (Table3). Ф is the 397 sediment correction factor of debris flow, which can be calculated by: (γ -γ ) 398 c m (10) ф= - (γs γc) 3 3 3 399 where γm is the density of water (t/m ) (1.00 t/m ); γs is the density of the solid material (t/m ) and usually 3 3 400 determined as 2.65 t/m ; and γc is the density of debris flow (t/m ). The γc listed in Table 2 was used for the 401 calculation. As the soil samples are not enough to cover all the 6 sections, the γc value at each section is selected 402 according to the nearby sample (Table 3). Wc is the area of debris flow cross section, which can be calculated by the 403 mud depth and terrain lines (Fig9 and Table 3).

404 The total volume of one debris flow (Qt) can be calculated by (Ou et al., 2006): 405 0.79 Qc=0.0188Qt (11) 406 The impact force of debris flow is the direct force that causes damage to prevention engineering and buildings. 407 The "8.20" debris flow had destroyed the no. 1 and 2 dam, the drainage channel in front of the no.1 dam, and two

408 bridges in the mizoguchi. Therefore, it is necessary to get the debris flow impact force (FC), which can help us 409 understand the cause of these damages. The FC can be calculated by (Ou et al., 2006): 410 (12) 훾푐 411 2 where λ is the form factor of building. Usually, λ Fis퐶 based= 휆 푔 on푉푐 푠푖푛훼the shape of the building: circular (1.0), rectangular 412 (1.33), square (1.47). α is the angle between the building surface and the direction of debris flow impact force (°). g 413 is the gravitational acceleration (9.8 m/s²). 414 Based on the above formulas, the dynamic parameters of the debris flow in different channel sections were 415 calculated and are listed in Table 5. It is worth noting that the amount of debris flow calculated by the 416 morphological survey method is closer to the measured value of 63.8×104 m3 than the value obtained by the rain 417 flood method. Because the rainfall data used in the rain flood method were measured from the position of the 418 mizoguchi, the actual rainfall in the formation region upstream of the ditch may be far greater than the rainfall data 419 we used. This result also confirms the reliability of the parameters such as the river flow rate, impact force and 420 debris flow rate calculated by the morphological survey method. 421 Table 5 Calculation results of related parameters Section number Calculation content Units C1 C2 C3 C4 C5 C6

Hc （m） 2.8 2.4 3.5 1.7 1.9 7.5

Bc （m） 46.5 45.8 76.3 81.3 67.9 106.2

Wc (m²) 128.54 202.86 329.01 217.37 166.48 459.81 J （‰） 185.29 80.1 141.28 54.59 132.41 113.15 16

γc （t/m³） 1.94 1.94 1.94 1.83 1.83 1.73 F （km²） 11.24 12.2 13.63 15.32 16.35 16.49 ψ / 2.54 2.62 2.79 2.87 2.99 3.12 τ h 2.79 2.90 3.09 3.19 3.32 3.49 S mm/h 17.8 17.8 17.8 17.8 17.8 17.8

nc / 0.27 0.07 0.10 0.08 0.03 0.05 ϕ / 1.32 1.32 1.32 1.01 1.01 0.79 L （km） 4.49 4.75 5.24 5.58 5.94 6.26

Dc / 2.5 2.3 2 1.9 1.5 1.2 λ / 1.33 1.33 1.33 1.33 1.33 1.33 α ° 90 90 90 90 90 90

Vc （m/s） 3.17 5.49 9.35 2.64 5.49 11.23

QP (m³/s) 32.86 35.34 38.68 43.19 45.46 45.02 Stormwater method Qc (m³/s) 190.912 188.8948 179.7803 165.1227 137.2116 96.89087 Qt (m³) 51775.33 51228.27 48756.41 44781.29 37211.78 26276.8

morphological Qc (m³/s) 407.1118 1114.649 3075.526 574.3737 914.795 5161.909 survey method Qt (m³) 55204.35 151146.4 417041.3 77885.07 124046.2 699954.8

FC （Pa） 26.40964 79.49356 230.0625 17.34015 74.9862 295.8936

422 4 Discussion 423 Eleven years after the “5.12” Wenchuan earthquake, such a large-scale debris flow can still erupt in the 424 earthquake zone, indicating that the region is still in the active period of earthquake geological disasters; the 425 reasons behind this activity should be the focus of our work. In this paper, we investigated the debris flow 426 disaster scene that occurred on August 20, 2019 and analysed the formation mechanism and dynamic 427 characteristics of this debris flow event by means of a drone survey, a high-definition remote sensing 428 interpretation and other means. However, at the same time, through this research, we also found many problems 429 that should be considered in future research; these issues mainly include the following two points. 430 (1) The acquisition of rainfall data in debris flow formation areas may not be representative. As one of the 431 conditions necessary for the formation of debris flows, rainfall is usually the most important inducing factor of 432 debris flows (Fan et al., 2019), so collecting rainfall data during debris flows should be the most important task 433 in research; however, this is not an easy task. The current rainfall monitoring stations are limited and are mostly 434 located in the lower Mizoguchi area. However, a large number of studies have shown that rainfall in debris flow 435 basins increases with elevation, so rainfall in debris flow formation areas is often much higher than in ditch 436 locations (Zhou et al., 2019). Our research results also prove this. Therefore, in future work, we should install 437 more rain collection instruments in key debris flow gullies, especially in debris flow formation areas, and avoid 438 using the rainfall data representing the ditch region to calculate dynamic parameters such as the flow rate of the 439 river and the flow rate of the debris flow. 440 (2) The provenance material amount may be underestimated. The current estimation of the provenance 441 material amount in the debris flow basin mainly determines the area by means of optical interpretation and then 442 estimates the volume using a statistical formula (Tang et al., 2011c). The problem with this method is that the 443 estimation of the movable depth of the provenance is often unable to reflect the specific starting depth, which

17

444 will lead us to underestimate the actual provenance. Through the research conducted in this paper, it was found 445 that the area of earthquake-induced loose deposits decreases sharply over time, but this does not mean that the 446 possible material provenance also decreases. The results of our study on the evolutionary characteristics of 447 provenance materials in multiple stages with different movement types show that during the 11 years after the 448 earthquake, the provenances continue to move through slide-flow and flow movements under the actions of 449 runoff and gravity. These moving sediments eventually settle thicker and thicker in the gradual areas and wait to 450 be restarted by rainfall. A more accurate provenance depth can only be determined by drilling or geophysical 451 analysis, and the depth that can be initiated may need to be further determined by measuring the shear strength 452 of the formation. In addition, the depth that can be initiated is also related to the actual volume of runoff. The 453 different flow strengths must be able to initiate the provenance at different depths. 454 In this paper, the calculation of the morphological investigation method compensates for the error caused 455 by the lack of rainfall data, but this calculation still cannot accurately indicate the current accurate provenance 456 volume. Although we have accurate provenance areas, we cannot determine the depth of all provenances. In 457 addition, because there are no field survey data or high-resolution images of the debris flow formation area, we 458 cannot analyse the actual start-up process of the debris flow in detail; thus, we need to supplement a detailed 459 investigation of the upstream channel and formation area in the next study.

460 5 Summary 461 Eleven years after the Wenchuan earthquake, a large-scale debris flow could still break out in Chediguan 462 gully in the earthquake zone, causing more serious losses than before. Studies of the cause and characteristics of 463 the "8.20" debris flow are of great significance for further understanding the status and activity of the debris 464 flow after the Wenchuan earthquake. Based on this concern, this paper investigated the Chediguan debris flow 465 disaster scene that occurred on August 20, 2019 and analysed the formation mechanism and dynamic 466 characteristics of this debris flow event by means of a drone survey, a high-definition remote sensing 467 interpretation and other means. 468 The main conclusions of this paper are as follows: 469 (1) The accumulated rainfall before the outbreak of the Chediguan debris flow at 2:00 on August 20, 2019, was 470 20.1 mm, and the accumulated rainfall in the previous period totalled 43.2 mm. The simulated rainfall that 471 eventually induced the debris flow appeared on the morning of August 20, 2019, from 2:00 to 3:00 in the 472 morning. The maximum rainfall intensity was only 17.8 mm, representing a slow-rising rain type. 473 (2) After the earthquake, from 2011 until August 2018, the number of provenances in the study area decreased 474 by 83 to a total of 123. The total provenance area increased from 473.8×104 m2 to 265.6×104 m2. The total 475 amount of provenance in the research area after the “5.12” earthquake showed a continuously decreasing trend, 476 but the provenances with different movement types showed different change rates. The total area of the 477 flow-moving provenances increased by 6.04% from 2011-2015, while the others decreased considerably. 478 Between 2015 and August 7, 2019, the total area of provenances moving as slide-flows, fall-slides, and flows 479 increased by 9%, 7.23% and 11.39%, respectively, while the total areas of provenances moving as slide-falls, 480 slides, and falls decreased by 14.06% and 30.79%, and 26.14%, respectively. After the debris flow broke on 481 20th Aug 2019, the number of provenances in the study area increased to 127. The total provenance area

18

482 increased to 281.2×104 m2. Among the flow types, the areas of provenances moving in slide-falls and flows 483 increased the most, with growth rates of 16.36% and 31.82%, respectively; these were also the main source 484 material of the "8.20" debris flow sediment supply. At present, the provenance in Chediguan gully is mainly 485 distributed in the main gully and moves in the form of slide-flows and flows. Therefore, a large amount of 486 dredging work in the main gully can effectively reduce the risk in the gully. 487 (3) The previous analysis of this paper shows that the “8.20” Chediguan debris flow event has the following 488 characteristics: 1) pre-rainfall brought the sediment together; 2) the flow cut down the groove, concentrating 489 the initiation of the provenance; 3) the main activity involved lifting erosion in the main channel; and 4) the 490 debris flow had an extremely high impact force. The Chediguan debris flow had an obvious chain effect, which 491 is characterized by the combination of heavy rain (mountain torrent)-collapse landslide-tributary debris 492 flow-rainfall-channel erosive erosion to form the debris flow. According to the parameter calculation, the 493 "8.20" Chediguan debris flow was a sub-viscous debris flow, the unit weight of the debris flow was 494 approximately 1.725 g/cm3, the maximum velocity at the mizoguchi was 11.23 m/s, the peak flow was m/s³, 495 the overall impact force of the debris flow at the gully mouth reached 295.9 Pa, and the total volume of the 496 debris flow exceeded 70×104 m3. The difference between the measured amount of outflowing debris (63.8×104 497 m3) and the calculated result was 1.55%, and the high coincidence between these values reflects the accuracy 498 of the calculated results in this paper. 499 The study shows that the impacts of the Wenchuan earthquake on the debris flow in the earthquake zone 500 still exist. The geological disasters in the Wenchuan earthquake zone are still in an active period, and 501 earthquake zone is still at a high risk for debris flows. Taking into account the causes and characteristics of 502 historic debris flows, detailed investigations and assessments of debris flows in the earthquake zone should be 503 carried out to execute the necessary dredging work for the watershed with excessive accumulation behind the 504 dam. In addition, it is important to carry out the necessary repair or rebuilding of damaged mitigation measures, 505 restore and enhance resilience to minimize the damage caused by debris flows, and establish a real-time 506 monitoring and early warning system for potentially high-risk debris flows to ensure early detection, early 507 prevention, and early management.

508 Acknowledgements 509 This research is financially supported by the National Key Research and Development Program of China 510 (Project No. 2017YFC1501004), National Natural Science Foundation of China (No. 41672299) and the 511 Scientific Research Foundation of Xihua University (Grant No. Z201133). The authors would like to thank the 512 State Key Laboratory for Geohazard Prevention and Environmental Protection (SKLGP), Chengdu, China, for 513 providing data and support with the fieldwork. Prof. Chuan Tang helped to improve the language and structure 514 of the paper. 515

516 References: 517 Chang M, Tang C, Van Asch T W J, et al. (2017). Hazard assessment of debris flows in the Wenchuan 518 earthquake-stricken area. South West China. Landslides, Vol 14, No. 5, pp. 1783–1792. DOI： 519 10.1007/s10346-017-0824-9. 520 Chang M, Tang C, Jiang Z L, et al. (2014). Dynamic Evolution Process of Sediment Supply for Debris Flow 521 Occurrence in Longchi of Dujiangyan, Wenchuan Earthquake Area. Mountain Research, (01):89~97. DOI: 522 10.3969/j.issn.1008-2786.2014.01.012. 523 Chen M , Tang C , Xiong J , et al. (2020). The long-term evolution of landslide activity near the epicentral area 19

524 of the 2008 Wenchuan earthquake in China[J]. Geomorphology, 367. 525 Dai F C, Xu C, Yao X, et al. (2011). Spatial distribution of landslides triggered by the 2008 Ms. 8.0 Wenchuan 526 earthquake, China. Journal of Asian Earth Sciences, Vol. 40 No. 4, pp. 883-895. DOI:10.1016/j.jseaes. 527 2010.04.01.0. 528 Fan R L , Zhang L M , Wang H J , et al. (2018). Evolution of debris flow activities in Gaojiagou Ravine during 529 2008–2016 after the Wenchuan earthquake[J]. Engineering Geology, 235:1-10. 530 Fan X M, Scaringi G, Domènech G, et al. (2019). Two multi-temporal datasets that track the enhanced 531 landsliding after the 2008 Wenchuan earthquake. Earth System Science Data, Vol. 11 No. 1, pp. 35-55. 532 DOI: 10.5194/essd-11-35-2019. 533 Fan X M, Scaringi G, Korup O, et al. (2019). Earthquake-Induced Chains of Geologic Hazards: Patterns, 534 Mechanisms, and Impacts. Reviews of geophysics. Vol.57 No. 2, pp. 421-503. DOI: 10.1029/2018RG000 535 626. 536 Ou G Q, You Y, Lv J, et al. (2006). A Study on the Scale Forecast and the Influence of Debris Flow on the 537 Water Diversion Project in the West Line Project of Water Diversion from Upper Yangtze River into 538 Upper Yellow River. Journal of Mountain Science, Vol. 5 No. 24, pp. 580-584. DOI: 539 10.1016/S1872-2040(06)60004-2. (In Chinese with English abstract). 540 Liu J F, You Y, Chen X, et al. (2014). Characteristics and hazard prediction of large-scale debris flow of 541 Xiaojia Gully in Yingxiu Town, Sichuan Province, China. Engineering Geology, Vol. 180, pp.55-67. 542 Liu J, You Y, Chen X , et al. (2015). Mitigation planning based on the prediction of river blocking by a typical 543 large-scale debris flow in the Wenchuan earthquake area[J]. Landslides, 13(5):1-12. 544 Tang C X, Van Westen C J, Tanyas H, et al. (2016). Analysing post-earthquake landslide activity using 545 multi-temporal landslide inventories near the epicentral area of the 2008 Wenchuan earthquake. Natural 546 Hazards and Earth System Sciences, Vol. 16 No.12, pp.2641-2655. DOI: 10.5194/nhess-16-2641-2016. 547 Tang C, Rengers N, Van Asch T W J, et al. (2011a) Triggering conditions and depositional characteristics of a 548 disastrous debris flow event in Zhouqu city, Gansu Province, northwestern China. Natural hazards and 549 earth system sciences. No.11, pp: 2903-2912. 550 Tang C, Zhu J. Qi X. (2011b). Landslide hazard assessment of the 2008 Wenchuan earthquake: a case study in 551 Beichuan area. Canadian Geotechnical Journal, Vol. 48 No. 1, pp. 128-145. 552 Tang C , Zhu J , Qi X , et al. (2011c) Landslides induced by the Wenchuan earthquake and the subsequent 553 strong rainfall event: A case study in the Beichuan area of China[J]. Engineering Geology, 554 122(1-2):22-33. 555 Tzimopoulos C , Evangelides C , Vrekos C , et al. (2018). Fuzzy Linear Regression of Rainfall-Altitude 556 Relationship[J]. Proceedings, 2(11). 557 Zhou W, Tang C. (2014). Rainfall thresholds for debris flow initiation in the Wenchuan earthquake-stricken 558 area, southwestern China. Vol. 11 No. 5, pp. 877 - 887. DOI: 10.1007/s10346-013-0421-5. 559 He Y H (2015). Research on the development characteristics and counter measurement of the debris flow in 560 Chediguan gully along the state road of G213. Southwest Jiaotong University. (In Chinese). 561 Tu G Q, Cheng Z L, Liu JK, et al. (2013). Assessment of Debris Flow Danger in High-intensity Earthquake 562 Area after Intense Earthquake—A Case Study of the Yingxiu-Chediguan Section of G213 Highway Hit 563 by the Wenchuan Earthquake. Journal of Mountain Science, Vol. 31 No. 04, pp. 391-398. 564 DOI:10.3969/j.issn. 008-2786.2013.04.002. (In Chinese). 565 YU B (2008). Research on the Calculating Density by the Deposit of Debris Flows. Acta sedimentologica 566 sinica, Vol. 26 No. 5, pp. 789-796. DOI: 10.1007/s11442-008-0201-7. (In Chinese). 567 Zhang W X (2016). The Analysis of Energy Dissipation Effect of Chediguan Gully on the Du Wen Highway.

20

568 Chengdu University of Technology. (In Chinese). 569 Zhou Q X, Kang L, Jiang XW, et al. (2019). Relationship between Heavy Rainfall and Altitude in 570 Mountainous Areas of Sichuan Basin. Meteorological Monthly, Vol. 45 No. 06, pp. 811-819. 571 DOI:10.7519/j.issn.1000-0526.2019.06.007. (In Chinese).

21